- Mar 24, 2013
- 1,787
- 95
- 91
GTX Titan Ultra is ready to launch. A full GK110 ready to make sure Nvidia have the fastest single GPU on the market.
Coming out 16/17th October.
http://wccftech.com/evidence-surfaces-nvidia-gk110-based-card-titan-ultra/
Titan Ultra will be based on Quadro K6000.
EDIT:
Nvidia will be launching Nvidia ATLAS on October 16/17th. It is GK180. It features 2880 cores instead of K20X with 2688 cores (GTX Titan), but will use the same power.
This could be our next GTX Titan Ultra ()
Looks like Nvidia have improved Kepler.
http://videocardz.com/46388/nvidia-launch-tesla-k40-atlas-gk180-gpu
Coming out 16/17th October.
The full blown GK110 Core optimized to gaming perfection is possible one of the most awaited chips of all time. Unlike the other current generation cores which suffer in Double Precision the GK110 core has the best of both worlds – heat output of current gen and double precision of Fermi. And of course not to mention that since the Titan is basically a slightly crippled Gk110 Core a full fledged GK110 core should return the figurative crown to Nvidia.
http://wccftech.com/evidence-surfaces-nvidia-gk110-based-card-titan-ultra/
Titan Ultra will be based on Quadro K6000.
EDIT:
Nvidia will be launching Nvidia ATLAS on October 16/17th. It is GK180. It features 2880 cores instead of K20X with 2688 cores (GTX Titan), but will use the same power.
This could be our next GTX Titan Ultra ()
Looks like Nvidia have improved Kepler.
The GK180 is not a typo, this has nothing to do with GK108. In fact this is a GK110 derivative with 2880 CUDA cores on board. Of course I’m just assuming GK180 came from GK110, but you never know. According to the leaker this is just GK110 with few minor changes.
The GK180 looks like a more binned GK110 chip, since even though it has more CUDAs, higher bandwidth and higher floating point performance, it still consumes just 235W (245 SXM), which is exactly the same TDP as for K20X. Speaking of performance this card will be slightly faster than K20X. With a peak single precision computing performance at 4.0 TFLOPS it’s just few percent faster than K20X, but the 4.0 TFLOPS is just preliminary number, it can go higher if the clock would be set even further.
The TESLA K20 ATLAS would feature 12GB memory. It would also feature GPU Boost technology known form GeForce series, although it would only work under specific workloads (AMBER, ANSYS).
http://videocardz.com/46388/nvidia-launch-tesla-k40-atlas-gk180-gpu
Last edited: